"Inspection passed at AQL 2.5" is not a guarantee that a shipment contains a particular percentage of defective goods. It records an acceptance decision under an agreed sampling plan. Buyers need the plan, the defect definitions and the actual inspection results to understand that decision.
What the system does
Checking every unit in a 10,000-piece order is impractical. AQL sampling checks a defined subset and uses the result to accept or reject the whole lot, with a known and stated risk of being wrong.
Three inputs determine the outcome:
- Lot size — how many units in the shipment.
- Inspection level and regime — select the level, normal/tightened/reduced inspection and single/double/multiple sampling scheme for the agreed product and standard edition.
- AQL and defect classes — the acceptance quality limit indexes the plan. Define which defects or nonconforming units are counted and how critical findings are handled.
Use the licensed tables for the exact agreed edition and plan to obtain the sample-size code and acceptance/rejection criteria. Single, double and multiple plans have different decision rules; do not infer the rule from the AQL value alone.
Reading a typical setup
Write the actual lot size, agreed standard edition, inspection level, sampling scheme, sample size and acceptance/rejection numbers on the inspection brief. The following defect descriptions are examples to discuss with the buyer, not a default sampling table:
| Defect class | Example concern | Decision to document |
|---|---|---|
| Critical | Safety or regulatory hazard | Define a separate rejection/escalation rule and required tests. No critical findings in a sample does not prove the whole lot is free of them. |
| Major | Function, fit or saleability affected | Agree the defect examples, AQL and the applicable plan’s acceptance/rejection rule. |
| Minor | A defined cosmetic deviation | Set limits with photographs or a sample; a cosmetic deviation may be major for some applications. |
The consequential number is the defect classification, not the AQL figure. Whether a scratch on a visible surface is major or minor decides whether a lot passes, and that classification is a negotiated definition rather than an objective fact.
The part buyers underestimate
AQL 2.5 for major defects does not mean 2.5% of your shipment is defective. It means the sampling plan is designed so that lots at around that defect rate will usually pass. The actual defect rate in a passing lot can be higher, and the plan carries a known probability of accepting a worse lot and of rejecting a good one.
Two practical consequences:
A passing inspection is not zero defects. Review the plan’s operating characteristics and consumer/producer risks against the product requirements. A tighter plan, additional process controls or an appropriate screening method may be needed; neither a higher inspection level nor 100% inspection automatically eliminates all risks.
A failing inspection is not a disaster. It means the lot was rejected under the agreed plan’s decision rule. Some multi-stage plans require a further sample when the first result falls between acceptance and rejection thresholds; that is not yet a rejection. Agree sorting, corrective action or re-inspection according to the contract and the recorded findings.
Defining defect classes before the inspection
This is where inspections go wrong, and it is entirely preventable. If "major" and "minor" are not defined in writing with reference to the golden sample, the inspector applies their own judgement and both parties are surprised.
A workable defect list is specific to the product and states, for each defect type, which class it falls into:
- Scratch longer than a stated length on a stated surface — major.
- Colour deviation beyond the approved swatch — major.
- Loose thread — minor.
- Missing safety label — critical.
Review the list with the supplier and inspector before production, and keep it attached to the approved specification.
What an inspection covers beyond defect counting
A competent pre-shipment inspection also verifies things that are not defects at all, and these frequently matter more:
- Quantity — cartons counted, contents verified against packing list.
- Dimensions and weight against specification.
- Packaging and shipping marks — often where errors are found.
- Function tests on a defined subset.
- Barcode scanning — a barcode that does not scan stops goods at the retailer.
Timing
Agree the production and packing condition with the inspector. Make the declared lot accessible for random selection, state what was unavailable and allow time for corrective action before loading. A packing percentage alone does not prove the inspection covered the whole order.
Consider an earlier process or production inspection where the product risks justify it. Its scope is different from final lot acceptance; confirm the actual production stage, available units and required checks before booking.
References and inspection scope
See the ISO 2859-1 overview and NIST explanation of acceptance sampling. Confirm the standard edition in your contract and use its licensed tables. Share the lot size, product specification and defect list to discuss an inspection brief.